I could discover that it fails when it reaches 4GB of memory alocation. > My production server with AAF (DRb) is suddenly (I can't precise when) > reaching 3GB+ of memory allocations, see:
More, apparently there is serious memory leaking with long queries with operators and wildcards. In my tests a single (a* OR e* OR i* OR o* OR u*) query made the memory allocation jumps from 30MB to 365MB immediately. Running the same query again the memory increased just a little bit. Running (b* OR c* OR d* OR f* OR g*) the increase was also incremental (small). Running (c* b* d* f* g*) it jumped to 412MB. Running (a* OR b* OR c* OR d*) in another (the ones before were all on the Post model, now we are searching the Blog model) model made it jump to 462MB. Running (a* OR b* OR c* OR d*) on the new model made it jump to 611MB. As we can see, its getting bigger and bigger and will soon reach the 4GB and fail. However I can't see what is really causing the error or where its happening. The unique message appears on the DRb server and is the [FATAL] failed to allocate memory. Any help is welcome!!! -- Posted via http://www.ruby-forum.com/. _______________________________________________ Ferret-talk mailing list [email protected] http://rubyforge.org/mailman/listinfo/ferret-talk
